Skip to main content

View Post [edit]

Poster: Kenneth Udut Date: Feb 14, 2016 10:00am
Forum: movies Subject: Re: Enough is DOUGH - //^\\

I'm going to look more into the wrapper, although I'm _almost_ done with this part of the project.

The keywords are actually distinct: I extracted the keywords from the hashtags I used to make the original posts on Vine (see https://vine.co/kenneth.udut for them in context) after downloading them with youtube-dl, which nicely used the description as the actual filename, including the hashtags, which saved me quite a lot of time.

Still, turning hashtags into subjects was surprisingly challenging - but that's because I'm using basic tools like Excel2k and Notepad++ for most of the text manipulation. [I automated some things with a little VBA code, like creating the directories and XML files]

I've only hacked around a little in Python, usually just enough to change some code I wanted different for a project, and I did a few fun things with a plugin for Minecraft that could read Python code and made it do a few things for me.

Still, if the solutions I'm putting together using my famliar old-school tools (keyword management and such) turns out to be too much for them, or if I can find some benefit with utilizing Python, I'll definitely make use of it.

Reply [edit]

Poster: Kenneth Udut Date: Feb 14, 2016 10:26am
Forum: movies Subject: Re: Enough is DOUGH - //^\\

Part of my search (and Jeff was nice enough to point me to this page for help in this: https://archive.org/browse.php?collection=kennethudut&;field=subject ) - is identifying the Hapax Legomenon that might be significant.

But first, they need to be collected and keywords processed.

Then the upper/lowercase ones normalized so that thing like Science and science and ScIeNcE show up in the same place.

Then getting rid of the mistakes I made.

From there, I want to discover themes that go beyond the hashtags extracted. There's no really good 'automatic categorization" tools that I've found yet that can work with short texts or small amounts of keywords to find out overall themes. Our AI capabilities are still primitive to where they'll be eventually.

I got _some_ help from http://act-dl.base-search.net/ - it helped me identify some themes (not of the Vines but of other writings - I threw 9000+ posts I made through the years through it and it spit out likely themes which were very helpful.

But still, not quite there yet.

I don't know what my end goal is. I somewhat envy people with definite purpose and intent in mind like Noah. I'm just floundering along from one project to another, hoping to discern the "who am I?" question as best I can.

Maybe there's a book in me. Maybe not. I don't know. But I _do_ know I have lots of raw data that can be processed in different ways, seen from different views and _maybe_... just maybe, if I can identify hidden patterns and start pulling things together (like these videos) into proper accurate categories beyond vague large categories like Science or Psychology or Motivation... *maybe* some intent and purpose can be driven by the data.

And then, perhaps I'll have something mind-blowingly awesome to contribute that even Noah would approve of.

Or maybe not. But I'm grateful that IA is supportive of my unorthodox process and willing to put up with my "baby steps" here.